<h2>Must know</h2>
<ul>
<li><strong>AI agent security vulnerabilities are a growing concern</strong>: Snyk’s research, ToxicSkills, uncovered critical security flaws in the AI Agent Skills ecosystem, with 13.4% of scanned skills containing severe issues and 76 confirmed malicious payloads designed for credential theft and data exfiltration. This highlights a significant supply chain risk for enterprises adopting AI agents. Operant AI launched Agent Protector to address these risks with real-time security for autonomous AI agents, emphasizing the need for platform providers to embed security rather than leaving it to customers. <a href="https://app.daily.dev/feed-by-ids?id=0BdgmYNfg&amp;id=0e28QTlYb&amp;id=2vNCw5dB4&amp;id=dL9ys7HAk&amp;id=the2EenMO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Valve delays hardware due to AI-driven component shortages</strong>: Valve has postponed pricing and shipping announcements for its Steam Machine and Steam Frame VR headset, citing severe memory and storage shortages. These delays are directly attributed to the surging demand from AI data centers, which has caused RAM prices to more than double since October 2025. This illustrates a tangible impact of the AI boom on the broader tech hardware supply chain, affecting even established gaming companies. <a href="https://app.daily.dev/posts/iXYPYa4qJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Global cyberespionage campaign targets critical infrastructure</strong>: A sophisticated nation-state threat group, TGR-STA-1030, has been conducting a global cyberespionage campaign since January 2024, compromising government and critical infrastructure in 37 countries. The group uses phishing, N-day exploits, and a new Linux kernel rootkit called ShadowGuard, which leverages eBPF to hide its activities. This campaign underscores the escalating threat landscape and the advanced techniques used by state-sponsored actors. <a href="https://app.daily.dev/posts/KNu0fQxJH" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Rust continues to gain traction in critical systems</strong>: WhatsApp rebuilt core media parsing components in Rust to enhance security for billions of users, demonstrating Rust’s growing adoption for client-side security at scale, while GNU Nettle 4.0 now includes SLH-DSA support, a post-quantum digital signature scheme, preparing systems for quantum-resistant security. <a href="https://app.daily.dev/feed-by-ids?id=6xPdh6fKm&amp;id=5YsTsMWni" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft appoints engineering quality czar</strong>: Microsoft CEO Satya Nadella appointed Charlie Bell as engineering quality czar, reporting directly to him, to address ongoing issues like Azure outages, Windows patch problems, and low Copilot adoption rates, with Hayete Gallot returning to lead security. <a href="https://app.daily.dev/posts/NGytU6zYE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS launches new EC2 instances</strong>: AWS introduced new EC2 High Memory U7i-6TB instances with custom Intel Xeon Scalable Processors for in-memory databases, and C8id, M8id, R8id instances powered by custom Intel Xeon 6 processors, offering significant performance and memory bandwidth improvements. <a href="https://app.daily.dev/posts/JPt4mjvME" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Substack discloses user data breach</strong>: Substack revealed a data breach from October 2025 that exposed user email addresses, phone numbers, and internal metadata, remaining undetected for months, though passwords and financial information were not compromised. <a href="https://app.daily.dev/posts/myrJIoXYs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Intel Arc B390 Linux performance analysis</strong>: Intel Arc B390 graphics with Xe3 architecture show competitive performance on Linux, but require specific configuration adjustments like performance mode and kernel 6.19+ to overcome default power limits and achieve optimal functionality. <a href="https://app.daily.dev/posts/GxsXcKIKI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ubuntu 24.04.4 LTS HWE update released</strong>: Ubuntu 24.04.4 LTS HWE update is now available, featuring Linux kernel 6.17 for improved power management and AMD hybrid laptop support, along with Mesa 25.2.7 for production-ready NVK and full Vulkan 1.4 compliance. <a href="https://app.daily.dev/posts/7U8PoA0ga" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GNOME Shell and Mutter 50 beta released</strong>: GNOME Shell and Mutter 50 beta releases introduce stable Variable Refresh Rate (VRR) support, improved frame scheduling, enhanced discrete GPU detection, and better virtual monitor and HiDPI handling for remote desktop scenarios. <a href="https://app.daily.dev/posts/buHOzLdcP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s super bowl ad stunt against openai</strong>: Anthropic is reportedly planning Super Bowl ads to mock OpenAI’s move into advertising, depicting chatbots turning into annoying salespeople. Because nothing says ‘we’re serious about AI for work and deep thinking’ like spending millions on a Super Bowl ad to dunk on your competitor. Meanwhile, OpenAI’s CEO is calling them ‘authoritarian’ for not wanting ads. It’s like watching two toddlers fight over a shiny new toy, but with billions of dollars at stake. <a href="https://app.daily.dev/posts/TmM2mNSSP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Elixir is the ‘best language for AI’ according to an unverified study</strong>: An article boldly claims Elixir is the ‘best language for AI,’ citing an unnamed ‘Tencent study’ that supposedly found Elixir achieved a 97.5% LLM completion rate. Apparently, its immutability and decade-long stability make it perfect for AI, unlike those other messy languages. It’s almost as if a language’s own community might be a <em>tad</em> biased when declaring it superior without any verifiable evidence. <a href="https://app.daily.dev/posts/1D7tGyRwk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>UK’s deepfake detection plan criticized as insufficient</strong>: The UK government announced plans to develop a framework for evaluating deepfake detection tech, partnering with Microsoft and academics. But cybersecurity expert Dr. Ilia Kolochenko basically said, ‘Bless your hearts, but we already have open-source tools for this, and the real problem is enforcement, not detection.’ So, while deepfakes surge from 500,000 to 8 million in two years, the government is busy evaluating the tools, not actually stopping the bad guys. Classic. <a href="https://app.daily.dev/posts/EQCYBDiDB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
